Beam Pattern Allocation Strategies for Satellite ... - IEEE Xplore

2 downloads 54 Views 427KB Size Report
Abstract—One of the major enabler for future wireless commu- nication technologies is spectrum sharing, allowing to effectively share spectrum portions among ...
IEEE ICC 2015 - Workshop on Cognitive Radios and Networks for Spectrum Coexistence of Satellite and Terrestrial Systems (CogRaN-Sat)

Beam Pattern Allocation Strategies for Satellite Cognitive Radio Systems Vincenzo Icolari, Daniele Tarchi, Alessandro Guidotti, Alessandro Vanelli Coralli Department of Electrical, Electronic and Information Engineering University of Bologna 40136 Bologna, Italy email:{vincenzo.icolari2,daniele.tarchi,a.guidotti,alessandro.vanelli}@unibo.it

Abstract—One of the major enabler for future wireless communication technologies is spectrum sharing, allowing to effectively share spectrum portions among different services. Even if much more considered in wireless terrestrial communications, spectrum sharing is increasing importance also in satellite communications where the spectrum scarcity issue is arising; cognitive techniques are a promising tool to cope with this problem. Taking into account the wide coverage and the flexibility that multibeam satellite systems can provide, in this paper we propose an optimization strategy that allows to efficiently assigns spectrum portions and polarizations to the beams. The proposed strategy is based on the definition of a Quality Index and aims at evaluating the beam pattern that maximizes the coverage. Assessments and evaluations of the potential gain in spectrum usage are performed with respect to the considered scenario.

I. I NTRODUCTION Requirements of future broadband services have lead to the necessity of designing new techniques for enabling wireless communication systems that fulfill such needs. Among others, one of the main issue is concerning the spectrum scarcity. Cognitive Radios (CR) allow to efficiently share spectrum portions while limiting harmful interference among different communication systems [1]. Recently, attention to CR techniques for spectrum coexistence in Satellite Communications (SatComs) [2] has also raised. However, due to their specific characteristics, such as the wide geographical satellite coverage, the design in SatComs scenarios of cognitive techniques for spectrum sharing and its efficient exploitation, relies on the flexibility that the cognitive satellite system could have. Next generation satellite systems are planned to be multibeam high throughput satellites (HTS) able to cover wide areas with multiple beams and to ensure broadband services provision to users with satisfactory QoS. Taking into account the increasing number of users, a flexible and reliable satellite system has to be designed for future purposes. With respect to future service requirements, a system design procedure for Terabit satellite systems deployment is given in [3], whereas an irregular beam size and non uniform bandwidth allocation for future unbalanced services demands with respect to the covered geographic area is proposed in [4]. However, exploitation of other bands already assigned to different users could provide the additional gain needed and, therefore, implementation of cognitive techniques in satellite systems might have a central role within this context.

978-1-4673-6305-1/15/$31.00 ©2015 IEEE

Furthermore, interest in deploying new broadband services in high frequency bands is arising. With particular focus on Ku- and Ka- bands, the non exclusive frequencies from 17.3 to 17.7 GHz and 17.7 and 19.7 GHz have been made available for deployment of the downlink of uncoordinated Fixed Satellite Services (FSS) by CEPT Decisions ERC/DEC/(05)08 and ERC/DEC/(00)07 respectively. However, these frequencies have already been allocated for Broadcasting Satellite System (BSS) feeder links and Fixed Services (FS) point-to-point backhauling links according to the provisions of the Radio Regulations [5]–[7]. In the depicted scenarios, the most harmful interference levels are those generated by incumbent users against FSS cognitive earth terminals, whereas FSS downlink transmissions are limited by Article S21 of the ITU Radio Regulations [6]. Specifically, the latter defines FSS emission limits in the frequency bands up to 40 GHz that prevents the incumbent user from being interfered. Within this context, one of the main challenges is the definition of bands that can be used by FSSs and the related transmission parameters (i.e., MODCOD, transmission power, etc.). Implementation of cognitive techniques for a flexible exploitation of satellite system resources with respect to the knowledge of incumbent users’ presence and activity, may lead to an overall system performance gain. The reuse of frequencies defined as non exclusive aims at a wider spectrum portion in addition to exclusive frequencies in which FSSs are usually allocated, even in presence of incumbent users. In particular, although incumbent users have without prejudice the right of the non exclusive bands use, they are located in limited areas or are very directive links, causing high interferences only in few regions and subsequent inefficient use of the spectrum [8], [9]. Among other system parameters, the overall bandwidth segmentation and the related polarizations can be considered as a first coarse radio resource allocation for increasing the overall efficiency in shared spectrum environments. The combination of one frequency interval and the related polarization is often referred as color. Due to the reduced number of incumbents in the area covered by the satellite, in this paper we propose to adjust the color beam reuse scheme of the satellite in order to avoid the same frequencies used by incumbent users. In particular, different system’s Quality Indexes (QI) are defined starting from the knowledge of the signal-to-noise

1652

IEEE ICC 2015 - Workshop on Cognitive Radios and Networks for Spectrum Coexistence of Satellite and Terrestrial Systems (CogRaN-Sat)

plus interference ratio (SINR) that all users or some of them experience. Thus, the Network Control Center (NCC) aware of the SINRs, is in charge of calculating the QI for the selection of the best color scheme. The QI gives a representation of which scheme provides the better performance in terms of coverage and offered capacity at system level. Hence, the paper is organized in the following way. Taking into account features and models similar to those proposed for future satellite systems deployment in [4] and [3], a more detailed description of the considered scenario and satellite system model are given in Section II. In Section III the proposed QIs are presented, along-with the proposed optimization strategy. Further, in Section IV numerical results are presented and compared, while conclusions and future works are included in Section V. II. S YSTEM M ODEL We consider a GEO multibeam satellite system operating in non exclusive bands. In particular, the cognitive satellite system is able to cover Europe with M beams characterized by a fixed width and each of which is associated to one color among those available in the Frequency Reuse Factor (FRF) scheme. The cognitive satellite system uses different sub-bands in order to minimize adjacent and co-channel interference between the closest beams and reuses them to cover the overall footprint. Each color represents a specific pair constituted by one sub-band and one polarization, if double polarization is considered. The system exploits non exclusive Ka frequency bands for downlink transmissions, besides the exclusive ones. Although the former bands are already assigned to incumbent users, they are located in limited zones within the satellite coverage area and, therefore, the cognitive satellite system has to exploit the same bands while limiting the mutual interference. The aim of the designed cognitive satellite system is to assign colors to those beams over the coverage area according to users’ reception performance. A typical scenario that might result is shown in Fig. 1. In the proposed scenario, we consider the spectrum portion of 800 MHz between 17.3 to 18.1 GHz in which incumbent BSS feeder uplinks (from 17.3 to 18.1 GHz) and Fixed Services (from 17.7 to 18.1 GHz) are already allocated [5]. These frequencies are in general used by incumbents without any time restriction for broadcasting or backhauling purposes. However, only limited areas near incumbent transmitters are highly interfered due to the directivity of these incumbent links. Moreover, propagation mechanisms between stations on the surface of the Earth, as reported in the Recommendation ITU-P 452-15 [10], allows to further reduce the mutual interference. Even though the hypothesis of a time continuous exploitation of the spectrum by incumbents, we can assume that only some carriers are occupied by them. The main purpose of a cognitive satellite system is to reuse the shared frequencies without interfering or being interfered. In the context of cognitive scenarios, beam-hopping has been proposed in a dual satellite cognitive scenario as an enabling

Fig. 1. Scenario Representation

technology [11]. The proposed technique exploits inactive beams of the incumbent satellite user. However, unlike the scenario considered in [11], presence of incumbent users in the scenario considered cannot be assumed temporary and different bands should be used in order to serve the cognitive earth terminals that are interfered. To this aim, thanks to the capability of the system to reuse different sub-bands and carriers, we propose as first solution in order to avoid bands occupied by incumbent users or highly interfered areas, to identify and cover with different colors regions in which incumbents are located. Moreover, in case of no available carriers in the non exclusive band, the cognitive satellite system can rely on the exclusive ones. In the following, we assume that the cognitive system is able to select the frequency reuse scheme according to the knowledge of incumbents presence and spectrum activities. To this aim, the cognitive system relies on spectrum awareness techniques. The most common techniques are the use of databases containing information on incumbent users and spectrum sensing techniques. However, in the proposed scenario, the cognitive transmission does not harm the incumbent reception. Thanks to this characteristic, we can also rely on the SINR values that users experience during transmission and they feed back to the NCC. Therefore, a first system set up phase can be performed in order to allow the satellite to broadcast a signal used by earth terminals for calculating the SINR in the selected carriers and, then, transmit feedback channel information to the NCC as described in [12]. Feedback information channel are already included in standards as for ACM (Adaptive Coding and Modulation) in [13], [14] although additional signalling fields may be necessary for this purpose. After the spectrum awareness phase, during which the NCC collects the SINR values of all the users or some of them, in a centralized manner it decides which color reuse scheme will provide the best performance by calculating the

1653

IEEE ICC 2015 - Workshop on Cognitive Radios and Networks for Spectrum Coexistence of Satellite and Terrestrial Systems (CogRaN-Sat)

proposed quality indexes as described in Section III. The selection and the subsequent adjustment of the frequency reuse scheme according to the QI, may be complex due to switching operation of the satellite payload. Therefore, this operation can be performed only if necessary in presence of highly interfering incumbent users.

Taking into account that each beam cannot have more than one color and that each beam has one color assigned, the P matrix is subject to the two following constraints: C X

C X M X

III. B EAM PATTERN S ELECTION S TRATEGIES In the proposed scenario, the deployment of a suitable beam pattern is the first step for allowing coexistence between incumbent and cognitive systems in the same bands. In this paper, we propose to allocate colors available in the frequency reuse scheme to the beams according to different system level metrics, i.e., Quality Indexes (QI). However, due to the additional presence of intra-system interference generated among the satellite beams, i.e., co-channel and adjacent channel interferences, we resort to a frequency reuse scheme that takes both the interference types into account. Therefore, in a first instance, we consider for system evaluation and assessments only those schemes that alternate colors in order to minimize intra-system interference. However, the proposed QIs are defined from a general point of view and can be extended also to a mixed beam schemes. Select the color scheme that gives the best performance in terms of offered capacity and coverage can be considered as an optimization problem. In [15], [16] authors introduced different methods and approaches for optimization problems, which are normally modeled as follows. x ¯ = max F (x) x

Subject to fk (x) ≤ 0

k = 0, 1, . . . , K

(1)

∀ j = 1, . . . , M

pi,j = 1

(4)

i=1

pi,j = M

(5)

i=1 j=1

The obtained model results in a very high number of possible patterns, increasing with the number of beams, even if colors are reduced in number. However, the number of available schemes could be reduced by limiting the set of allowed schemes to those that minimize the intra-system interference. Thus, we can define a set P, which includes only those schemes that allocate same colors in non adjacent beams. Beginning from here we restrict our analysis to those patterns P belonging to the set P, i.e., to P ∈ P. Considering that more users can be served within the same beam, we define a matrix QP B constituted by the elements Qcm , which define the QI metric of the m-th beam to which is assigned the c-th color. For a certain scheme P we obtain:   1 Q1 · · · Q1M  . ..  ..   (6) QP . B =  .. .  C C Q1 · · · QM (C×M ) From the weighted-sum method, Qcm is defined as linear combination of the SINRs of the users in the same beam and can be expressed as:

(2)

In equations 1 and 2 are defined i) the vector of the problem variables x, ii) F ( · ) and iii) fk ( · ), respectively, the objective and the set of constraint functions. Aim of the optimization problem is to maximize the selected objective function F ( · ) finding x ¯. In this paper, we consider the weighted-sum method to model the optimization problem and the related utility function [15]. The utility function, which attempts to model the decision preferences, can also be named Quality Index metric since it usually represents the experienced satisfaction. Thus, we define QP and Qcm , respectively, as the utility functions to be maximized of the considered combination scheme and of the beam, that is at the base of the color combination selection strategy, by considering the m-th beam and the c-th color. We assume that the NCC is able to evaluate the QI by linearly combining SINRs of each user when the color scheme P is considered. Colors can be assigned in different combinations to beams, thus generating different schemes, which can be described by a matrix P in which each element pi,j equal to one denotes the i-th color assigned to the j-th beam.   p1,1 · · · p1,M  . ..  ..  P= (3) . .   .. pC,M · · · pC,1 (C×M )

Qcm =

N X

wnc SIN Rnc

withc = 1, . . . , C

(7)

n=1

where N is the number of selected users used for calculating the SINR values1 , and c is the color assigned to the m-th beam among those in the frequency reuse plan. Therefore, SIN Rnc represents the SINR value experienced by the n-th user if the c-th color is assigned to the m-th beam, whereas wnc the respective weight. Define suitable weights in order to give a reliable representation of the performance of the selected scheme has central importance. Due to the interference that incumbents cause, the cognitive user might not be able to exploit the same bands. Therefore, if an incumbent is present, some users located in the same area can not be reached by the transmitted signal preventing the satellite to close the link budget. In order to avoid this issue, a different band should be selected. Since incumbents’ presence damages only a limited number of users within the beam, we can consider the maximization of the number of users that can be served as the main objective of the scheme selection strategy. Here, two different approaches are therefore proposed. After defining a suitable threshold γth 1 The users selected for evaluating the SINR could be a subset of all the users within a certain beam [17].

1654

IEEE ICC 2015 - Workshop on Cognitive Radios and Networks for Spectrum Coexistence of Satellite and Terrestrial Systems (CogRaN-Sat)

TABLE I L OSSES OF THE DIFFERENT FREQUENCY REUSE SCHEMES . Pattern 3S1P 2S2P 3S2P 4S1P

Maximum Coverage Losses %

Maximum Capacity Losses [Mbit/s]

1,58 3,70 2,39 2,26

4 389 7 574 4 620 3 258

IV. N UMERICAL R ESULTS

Fig. 2. Set of possible colors, defined as combination of frequency segmentation and polarization.

that determines if a user can be served or not when compared to SIN Rnc , a proper weight wnc for the QI calculation can be defined as: ( c w1,n

=

1 c SIN Rn

if SIN Rnc ≥ γth

0

if SIN Rnc < γth

(8)

By using this approach, it is possible to identify how many users are above the threshold and choosing the one that maximize this number. The latter proposed approach considers SINR values weighted linearly with respect to the threshold. Thus, we obtain: c w2,n =

SIN Rcn − γth γth

(9)

Hence, users that experience a higher SINR will be also weighted more. By means of this second approach, users with good performance tend to encourage the selection of the color that provides them the best performance. By considering uniform coverage and service demand over the whole area, we assume that beams have equal service demands. Thus, we define the overall system QI QP as the average between the M Qcm values of the matrix QP B assigned to the considered scheme P :

QP =

C M 1 XX c Q , M c=1 1 m

Qcm ∈ QP B

(10)

Finally, the maximization problem can be defined as ¯ P = max{QP (SIN Rc )} Q n P ∈P

(11)

and subject to considering only patterns of the set P, and a threshold γth .

Analysis and assessments of the frequency reuse scheme and color combinations that provide best performance in terms of coverage and offered capacity in the proposed scenario, have been performed. Incumbent users are placed randomly within the coverage area of the satellite, whereas their power levels are uniformly generated starting from typical values [8]. For simplicity, the incumbent users’ signal interferes a uniform area around the source. In addition, assuming continuous usage of the spectrum by incumbent users, we can take into account that the interference generated against cognitive FSS earth terminals is constant in time and occupies a wide band up to 400 MHz and can affect both polarizations. The cognitive satellite system exploits the band between 17.3 GHz and 18.1 GHz, which provides 800 MHz. A number of colors C equal to 3, 4 and 6 in different polarizations are considered. Therefore, several possible frequency reuse schemes are developed, as shown in Fig. 2. These schemes are identified with the abbreviation ‘aSbP’, in which a represents the number of sub-bands S and b the polarizations P. Consequently, the number of colors is the product ab and to each beam can be assigned a sub-band polarization pair among those of the frequency reuse scheme. The width granted for each sub band will be equal to the total forward bandwidth over the number of sub bands per polarization. Computer simulations have been performed at sub-band level. Moreover, among the 200 beams used to cover Europe, inter-beam interference is considered apart from the interference generated by incumbent users. The forward link of the cognitive satellite system is based on a DVB-S2 transmission with ACM and the transmission is considered reliable if an availability equal to 99.7% is guaranteed. The set P of the considered color combinations for each frequency reuse scheme, includes only those that minimise the intra-system interference as described in Section III. The different frequency schemes and the respective color combinations are evaluated in terms of coverage and total offered capacity. In particular, the coverage represents the percentage of covered area for which an availability equal to the 99.7% is reached whereas we assess the total capacity that the system can offer. In Fig. 3 and 4 the minimum and the maximum values of the coverage and of the system capacity obtained for different combinations of the considered frequency reuse schemes among those proposed in Fig. 2, are shown. Results are also compared with an equivalent scenario in which the cognitive

1655

IEEE ICC 2015 - Workshop on Cognitive Radios and Networks for Spectrum Coexistence of Satellite and Terrestrial Systems (CogRaN-Sat)

Fig. 3. Coverage percentages for the different FRF considered.

c Fig. 5. Normalised QP function of the number of users with w1,n weights.

Fig. 4. Offered Capacity for the different FRF considered.

c Fig. 6. Normalised QP function of the number of users with w2,n weights.

satellite operates in the same frequency bands in case of incumbent users’ absence. The percentages of the coverage achieved are shown in Fig. 3 where, as expected, a higher number of sub bands provides a higher coverage since colors interfere less each other and more combinations are possible. However, a higher number of sub bands lead to a narrower bandwidth per beam and thus to a lower capacity. This is shown in Fig. 4 where it is confirmed that the four color reuse scheme in single polarization reaches a lower capacity. The same trends are foreseen also in case of no incumbent presence since the same system parameters are considered for all the frequency reuse schemes. With respect to the scenario with presence of incumbent users, the frequency reuse scheme that suffers most losses in terms of coverage and capacity is the ‘2S2P’ scheme whereas the others have similar losses as highlighted in Tab. I. Moreover, in Fig. 3 and 4 the minimum and the maximum values obtained for different combinations of the color schemes considered, slightly differ each other. This is due to the constraint of assigning different colors to adjacent beams. Therefore, under these assumptions we cannot avoid incumbent users but only mitigate the harmful interference they cause. However, it is possible to identify the best color reuse scheme among those available.

Due to the low number of allowed combinations, the best one can be identified calculating the proposed QP from the c matrix QP B : weights are derived and associated to the SIN Rn values of the users known at the NCC. Thus, the respective QI metrics as described in Section III are calculated. Results are shown with respect to the frequency scheme ‘4S1P’, which provides the best performance in terms of coverage. Fig. 5 and 6 show, respectively, the results obtained for the QI calculation of different combinations of the same color scheme, i.e., ‘P1’ and ‘P2’ are different color combinations c c and w2,n of the ‘4S1P’ frequency reuse scheme, when w1,n are considered. The threshold γth is fixed to -2 dB, that is the SINR value proposed in [13] as the minimum quality-ofreception value at the receiver. This value can represent the threshold between typical SINR values in case of incumbent presence or absence. Thus, QP is calculated as function of the number of users for all the possible pattern combinations. Users are randomly selected within the beam and their SINR values collected at the NCC. Results show that the scheme that provides the best performance in terms of coverage and capacity achieves the highest QP allowing the selection of the best combination among those available within the set P. However, in case of a limited number of users, the QP may not be reliable since few samples are considered and this can

1656

IEEE ICC 2015 - Workshop on Cognitive Radios and Networks for Spectrum Coexistence of Satellite and Terrestrial Systems (CogRaN-Sat)

be noted in case of N = 5 in Fig. 5. The greater the number of users, the more reliable would be the color combination selection since the QP will be sum of a higher number of users’ samples randomly selected around the beam. However, due to the slight difference between the combinations in terms of performance, this trend is identified also in the QP calculation. V. C ONCLUSIONS Spectrum sharing is an effective tool for improving the usage of the spectrum. In the scenario considered a cognitive satellite system is able to operate in the same frequencies already allocated to BSS feeder stations and FS links without interfering them and allowing to gain spectrum for service provision. However, presence of incumbent systems prevents users close to them to be reached by service provision. Due to the high flexibility that the next generation satellite systems can provide, in this paper different frequency reuse schemes have been evaluated. Moreover, an approach based on the SINR values fed back by the users in order to select the color combination scheme that provides the highest coverage, is proposed. Considering only combinations in which adjacent beams are not assigned to same colors in order to minimize the intra-system interference, we reduce to a limited number the set of those available. The Quality Index defined in Section III in order to select among these schemes the most reliable one, yields to the best scheme in terms of coverage. However, an increasing number of users per beam has to be considered in order to give a true qualitative evaluation of the best pattern.

[9] “Adaptation and design of cognitive techniques for satellite communications,” Deliverable D3.3, 2014. [Online]. Available: http://www.ict-corasat.eu/ [10] Prediction procedure for the evaluation of interference between stations on the surface of the Earth at frequencies above about 0.1 GHz, ITU-R Recommendation P.452-15, Sep. 2013. [11] S. Sharma, S. Chatzinotas, and B. Ottersten, “Cognitive beamhopping for spectral coexistence of multibeam satellites,” in Future Network and Mobile Summit (FutureNetworkSummit), 2013, July 2013, pp. 1–10. [12] V. Icolari, A. Guidotti, D. Tarchi, and A. Vanelli-Coralli, “An interference estimation technique for satellite cognitive radio systems,” in Proc. of IEEE ICC 2015, London, UK, Jun. 2015, accepted for publication. [13] Digital Video Broadcasting (DVB); Second generation framing structure, channel coding and modulation systems for Broadcasting, Interactive Services, News Gathering and other broadband satellite applications (DVB-S2), ETSI European Standard EN 302 307, Mar. 2013. [14] Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 1: Overview and System Level specification, ETSI European Standard EN 101 545, Apr. 2014. [15] R. Marler and J. Arora, “Survey of multi-objective optimization methods for engineering,” Structural and Multidisciplinary Optimization, vol. 26, no. 6, pp. 369–395, 2004. [Online]. Available: http: //dx.doi.org/10.1007/s00158-003-0368-6 [16] J. Arora, O. Elwakeil, A. Chahande, and C. Hsieh, “Global optimization methods for engineering applications: A review,” Structural optimization, vol. 9, no. 3-4, pp. 137–159, 1995. [Online]. Available: http://dx.doi.org/10.1007/BF01743964 [17] V. Icolari, D. Tarchi, A. Vanelli-Coralli, and M. Vincenzi, “An energy detector based radio environment mapping technique for cognitive satellite systems,” in Proc. of IEEE Globecom 2014, Austin, TX, USA, Dec. 2014.

ACKNOWLEDGMENT This work was partially supported supported by the EU FP7 project CoRaSat (FP7 ICT STREP Grant Agreement n. 316779). R EFERENCES [1] I. F. Akyildiz, W.-Y. Lee, M. C. Vuran, and S. Mohanty, “Next generation/dynamic spectrum access/cognitive radio wireless networks: A survey,” Comput. Netw., vol. 50, no. 13, pp. 2127–2159, Sep. 2006. [2] K. Liolis, G. Schlueter, J. Krause, F. Zimmer, L. Combelles, J. Grotz, S. Chatzinotas, B. Evans, A. Guidotti, D. Tarchi, and A. Vanelli-Coralli, “Cognitive radio scenarios for satellite communications: The CoRaSat approach,” in Proc. of 2013 Future Network and Mobile Summit, Lisbon, Portugal, Jul. 2013. [3] A. Kyrgiazos, B. Evans, P. Thompson, P. T. Mathiopoulos, and S. Papaharalabos, “A terabit/second satellite system for European broadband access: a feasibility study,” Int. J. Satell. Commun. Network, vol. 32, no. 2, pp. 63–92, 2014. [4] A. Kyrgiazos, B. Evans, and P. Thompson, “Irregular beam sizes and non-uniform bandwidth allocation in hts,” in Proc. of 31st AIAA International Communications Satellite Systems Conference, 2014. [5] The European table of frequency allocations and applications in the frequency range 8.3 kHz to 3000 GHz (ECA table), ECC/CEPT ERC Report 25, Oct. 2013. [Online]. Available: http: //www.erodocdb.dk/Docs/doc98/official/pdf/ercrep025.pdf [6] Radio Regulations - Volume 1: Articles, International Telecommunications Union - Radiocommunication Sector Std., 2012. [7] Radio Regulations - Volume 2: Appendices, International Telecommunications Union - Radiocommunication Sector Std., 2012. [8] “Performance evaluation of existing cognitive techniques in satellite context,” Deliverable D3.2, 2014. [Online]. Available: http://www. ict-corasat.eu/

1657

Suggest Documents